Mitochondrial fatty acid synthesis (mtFAS) is a highly conserved pathway essential for mitochondrial biogenesis. The mtFAS process is required for mitochondrial respiratory chain assembly and function, synthesis of the lipoic acid cofactor indispensable for the function of several mitochondrial enzyme complexes and essential for embryonic development in mice. Mutations in human mtFAS have been reported to lead to neurodegenerative disease. The source of malonyl-CoA for mtFAS in mammals has remained unclear. We report the identification of a conserved vertebrate mitochondrial isoform of ACC1 expressed from an ACACA transcript splicing variant. A specific knockdown (KD) of the corresponding transcript in mouse cells, or CRISPR/Cas9-mediated inactivation of the putative mitochondrial targeting sequence in human cells, leads to decreased lipoylation and mitochondrial fragmentation. Simultaneous KD of ACSF3, encoding a mitochondrial malonyl-CoA synthetase previously implicated in the mtFAS process, resulted in almost complete ablation of protein lipoylation, indicating that these enzymes have a redundant function in mtFAS. The discovery of a mitochondrial isoform of ACC1 required for lipoic acid synthesis has intriguing consequences for our understanding of mitochondrial disorders, metabolic regulation of mitochondrial biogenesis and cancer.

Introduction

Acetyl-coA carboxylase (ACC) catalyzes the first committed step of the fatty acid synthesis (FAS) process, producing the malonyl-CoA extender for fatty acyl group elongation by the addition of a carboxyl moiety from carbonate to acetyl-CoA [1]. ACCs are highly conserved in organisms from archaea to higher eukaryotes [2]. In contrast with their bacterial counterparts, the eukaryotic ACCs are multifunctional enzymes, harboring all necessary activities required for the carboxylation reaction on a single polypeptide [1]. Mammalian ACCs have been reported to exist either as soluble cytosolic variants (ACC1 in humans, encoded by the ACACA gene) serving as a producer of malonyl-CoA for cytosolic FAS, or anchored to the cytoplasmic side of the outer mitochondrial membrane (ACC2 encoded by ACACB), where ACC contributes to the regulation of mitochondrial β-oxidation by controlling the rate of acyl group transfer into mitochondria [3]. Thus, malonyl-CoA fulfills several additional roles within cells apart from acting as an FAS extender by serving as a master regulator in a wide array of physiological processes linked to lipid metabolism. These include control of mitochondrial acetyl-CoA generation from fatty acids, the ketogenic response of the liver during fasting, balancing cardiac fuel usage and adjusting energy source selection in skeletal muscle [4].

It has been firmly established that mitochondria host a highly conserved FAS pathway (mitochondrial fatty acid synthesis, mtFAS). MtFAS follows the bacterial type II FAS mode [5], where the acyl carrier protein (ACP)-dependent synthesis of fatty acids is carried out by a series of discrete polypeptides catalyzing the individual reactions [6]. The requirement of a separate mtFAS pathway, conserved from yeast to higher eukaryotes including humans, indicates an essential role of this process in mitochondrial function.

In Saccharomyces cerevisiae, all mtFAS components are essential for respiration. A complete set of enzymes of this pathway required to produce straight chain fatty acids has been identified in yeast, and functional mammalian orthologs of nearly all of the yeast mtFAS proteins have been described [6]. Radioisotope labeling incorporation study in mammalian heart and plant has shown that the major mtFAS product is octanoyl-ACP, providing the precursor for endogenous synthesis of the lipoic acid cofactor essential for the catalytic activities of several key mitochondrial oxidative decarboxylation enzyme complexes. In mammals, pyruvate dehydrogenase, α-ketoglutarate dehydrogenase, branched-chain ketoacid dehydrogenase, α-ketoadipate dehydrogenase and the glycine cleavage system depend on lipoic acid for their function [79]. The mtFAS pathway is capable of synthesizing fatty acids longer than eight carbons, and there is evidence for an important role for mtFAS products longer than C8 in mitochondrial biogenesis [10].

The first cases of a novel neurodegenerative human disorder (MEPAN, mitochondrial enoyl-CoA reductase protein-associated neurodegeneration) caused by recessive mutations in the MECR (mitochondrial enoyl-CoA/ACP reductase) protein of mtFAS have been reported recently [11]. Owing to the large number of components involved in the mtFAS process, the identification of more cases of disease caused by mtFAS dysfunction is to be expected.

One fundamental aspect for which mammalian mtFAS has been described to deviate from the yeast mtFAS pathway is the generation of malonyl-CoA. Both mammalian ACC variants have been reported to contribute to the malonyl-CoA production in the extra-mitochondrial space, but not in the mitochondrial matrix, where the S. cerevisiae mtFAS ACC encoded by the non-canonically translated HFA1 gene is localized [12,13]. It has been suggested that mitochondrial propionyl-CoA carboxylase may be capable of producing malonyl-CoA in this compartment in mammals by using the much less preferred acetyl-CoA as a substrate [14]. An important contribution to solving the question on the origin of mitochondrial malonyl-CoA for mtFAS was provided by Witkowski et al. [9], who reported that the mitochondrial ACSF3 malonyl-CoA synthase found in animals could fulfill such a function. However, the extent of contribution of this protein to mtFAS is not clear, as a knockdown (KD) of ACSF3 expression in human cultured cells did not yield in decreased protein lipoylation, which can be observed in studies on the effects of KD of other mammalian mtFAS components [15,16] and in fibroblasts obtained from MEPAN patients [11]. Therefore, an alternative source may be postulated for malonyl-CoA production in mammalian mitochondria.

Here, we demonstrate the existence of highly conserved mammalian mitochondria-specific ACC1 isoforms required for lipoic acid synthesis and the maintenance of tubular mitochondrial morphology in mammalian cultured cells. Our data indicate that mitochondrially targeted ACC1 and ACSF3 act in concert to provide malonyl-CoA in mitochondria. These results add a final piece to the mtFAS puzzle and have intriguing consequences to our understanding of regulation of mitochondrial function and energy metabolism in health and disease.

Experimental

Bioinformatics

The mouse and human ACACA/Acaca sequence data were obtained from Emsembl (http://www.ensembl.org). Of the 23 annotated ACACA transcripts, only the four variants encoding full-size ACC1 were analyzed (Figure 1). The alignment of the transcript variants of the human ACACA was produced using the ClustalW program (http://www.ebi.ac.uk/Tools/msa/clustalw2/) (Figure 1A) [17]. Seven transcripts of mouse Acaca are annotated in the Emsembl database. The information on the two transcripts encoding full-size mouse Acc1 as well as an incomplete 5′-RACE-generated transcript is shown in the table in Supplementary Figure S1B. A transcript encoding a full-size mouse Acc1 with a predicted mitochondrial import sequence (Acaca-202; protein ENSMUSP0000048865; mitochondrial import probability 0.8676) had been previously present in ENSEMBL, but is currently not available anymore. For both human and mouse ACACA/Acaca transcript variants, the mitochondrial localization prediction for the encoded proteins was carried out using MitoProtII [18], TargetP [19] and iPSORT [20].

The N-terminal extension of human ACC1 isoform 1 constitutes a MTS.

Figure 1.
The N-terminal extension of human ACC1 isoform 1 constitutes a MTS.

(A) Alignment of the predicted protein products of ACACA transcripts encoding full-size human ACC1. The N-terminal extension is marked by gray shading, and the predicted MTS up to the signal peptide cleavage site is shown underlined and in bold letters. (B) Transcript IDs, length of predicted proteins and MitoprotII, TargetP and iPSORT for mitochondrial import predictions for the polypeptides. NP: not predicted. (C and D) Fluorescence microscopy analysis of HEK293 cells expressing EGFP (C) or EGFP N-terminally appended with the N-terminal extension of human ACC1 isoform 1 shown in gray background in A (D). To visualize the mitochondrial network, cells were co-transfected with the MtDsRED plasmid (red fluorescence panel). The fluorescence of EGFP lacking the N-terminal extension shows no correlation with the MtDsRED signal (C), while the N-terminally appended variant co-localizes with the mitochondrial stain (D). Scale bars are 10 µm. (E) Analysis of overlap (Manders coefficient, (1)) and correlation (Pearson's coefficient (2)) of the MtDsRED signal with control EGFP versus hACC1-EGFP fluorescence. The hACC-EGFP signal shows a higher overlap with the red fluorescence than the control construct and high positive correlation, while the EGFP control is negatively correlated with the mitochondrial fluorescence of the MtDsRED construct. Data were collected from the 3D representations that were created from eight individual cells. Student's t-test was performed. Errors bar indicate the standard deviation (Manders P-value: 0.0006; Pearson's P-value: 0.0001 (E)).

Figure 1.
The N-terminal extension of human ACC1 isoform 1 constitutes a MTS.

(A) Alignment of the predicted protein products of ACACA transcripts encoding full-size human ACC1. The N-terminal extension is marked by gray shading, and the predicted MTS up to the signal peptide cleavage site is shown underlined and in bold letters. (B) Transcript IDs, length of predicted proteins and MitoprotII, TargetP and iPSORT for mitochondrial import predictions for the polypeptides. NP: not predicted. (C and D) Fluorescence microscopy analysis of HEK293 cells expressing EGFP (C) or EGFP N-terminally appended with the N-terminal extension of human ACC1 isoform 1 shown in gray background in A (D). To visualize the mitochondrial network, cells were co-transfected with the MtDsRED plasmid (red fluorescence panel). The fluorescence of EGFP lacking the N-terminal extension shows no correlation with the MtDsRED signal (C), while the N-terminally appended variant co-localizes with the mitochondrial stain (D). Scale bars are 10 µm. (E) Analysis of overlap (Manders coefficient, (1)) and correlation (Pearson's coefficient (2)) of the MtDsRED signal with control EGFP versus hACC1-EGFP fluorescence. The hACC-EGFP signal shows a higher overlap with the red fluorescence than the control construct and high positive correlation, while the EGFP control is negatively correlated with the mitochondrial fluorescence of the MtDsRED construct. Data were collected from the 3D representations that were created from eight individual cells. Student's t-test was performed. Errors bar indicate the standard deviation (Manders P-value: 0.0006; Pearson's P-value: 0.0001 (E)).

Plasmids, cell lines, culture media and growth conditions

The oligonucleotide sequences used for plasmid construction and analyses in our study will be made available on request. The pEGFP-N1 vector (Invitrogen, Carlsbad, CA, U.S.A.; GenBank accession #U55762) was provided by Dr Sakari Kellokumpu (University of Oulu, Oulu, Finland). A sequence containing the first 513 nucleotides of the ACC1 ORF encoded by transcript variant ACACA-001 (ENST00000616317.4) was synthesized by GenScript (Jiangning, Nanjing, Jiangsu Province, China). A PCR product containing the first 111 nucleotides of this ORF, encoding the 37 amino acid N-terminal extension of ACC1 isoform 1 predicted to constitute an MTS (mitochondrial targeting signal), flanked by HindIII and PstI sites, was cloned into vector pEGFP-N to generate an in-frame fusion of the N-terminal region of ACC1 harboring the putative MTS (MWWSTLMSILRARSFWKWISTQTVRIIRAVRAHFGGI) to EGFP. For the oligonucleotide design for mouse KD studies, the sequences of the two isoforms of Acaca/Acc1 from mouse are referred to (Acaca-001 and Acaca-005, Supplementary Figure S1). Three KD cassettes targeted the homologous sequence shared by cytosolic and putative mitochondrial transcript variants (A1, A2 and A3), while two other cassettes targeted the exon predicted to encode the mouse Acc1 isoform 1 MTS (M1 and M2). Plasmids and methods used for the KD cassette generation have been described recently [21]. Briefly, for the shRNA KD of the ACC1 in mouse NIH3T3 cell lines, pRVH1-puro [22], pVSV and pSUPER [23] were provided by Aki Manninen (Biocenter Oulu, University of Oulu, Oulu). Oligonucleotides encoding shRNAs were cloned into pSUPER according to Brummelkamp et al. [23]. The shRNA expression cassette was then transferred into the XhoI/EcoRI site of the pRVH1-puro. Mouse 3T3 cell lines stably expressing Acaca KD cassettes were generated by retroviral transduction. The human Phoenix gag-pol packaging cell line (American Type Culture Collection with authorization by Garry Nolan, School of Medicine, Stanford University, Stanford, CA) was provided us by Aki Manninen (Biocenter Oulu, University of Oulu, Oulu, Finland). The construct used for KD of expression of the mouse mtFAS Mecr has been described [21]. All cell lines were maintained at 5% CO2 at 37°C in Dulbecco's Modified Eagle Medium (DMEM, 4.5 g/l glucose; Invitrogen, Carlsbad, CA, U.S.A.) supplemented with 10% fetal bovine serum (FBS) (HyClone, Cramlington, U.K.) and GlutaMAX (ThermoFisher Scientific, Waltham, MA, U.S.A.). For the investigation of functionality of the putative MTS constituted by N-terminal sequence of the human ACC1, the HEK293T (human embryonic kidney) cell line was used. For transfection, 150 000 cells were seeded in each well of a 6-well plate and incubated at 37°C. After 24 h, 1 µg of plasmid, 97 µl of Opti-MEM (Invitrogen, Carlsbad, CA, U.S.A.) and 3 µl of FuGENE 6 (Roche Diagnostics, Indianapolis, IN, U.S.A.) were added. At 72 h post transfection, the cells were fixed with 4% paraformaldehyde (PFA) for slide preparation. VECTASHIELD HardSet Mounting Medium with DAPI (VECTOR, CA, U.S.A.) was used.

Generation of Acaca and Mecr shRNA KDs

Acaca KD

As a control, empty vector RVH1 was used for the transfection. The Acaca-005 sequence (ENSMUST00000133811.2) was used to design the KD cassettes targeted to the predicted MTS. This transcript variant is annotated as incomplete and was identified by 5′-RACE [24]. As the overall structure of the mouse Acaca gene would allow to produce a full-size mouse Acc1 harboring this N-terminal extension, we chose to target this sequence for our KD studies. In 2011, ENSEMBL annotated a transcript Acaca-202, which showed a highly homologous sequence with Acaca-005 coding for a protein harboring the MTS and of a total length of 2382 amino acids. This transcript is currently not available on the ENSEMBL database anymore. A detailed description of the KD procedure can be found in Supplementary Experimental section. After the selection for successfully infected cells, protein and RNA were isolated for analysis and real-time quantitative PCR.

Mecr KD

Mecr shRNA KD cell lines were generated using the NIH3T3 cell line as described recently [21]. Cell lines were maintained as above, supplemented with penicillin–streptomycin (Sigma–Aldrich, Saint Louis, MO, U.S.A.).

Generation of the ACSF3 esiRNA KD

HEK293 cells were maintained in DMEM supplemented with 10% FBS at 37°C in a 5% CO2 atmosphere. Mission esiRNA reagents, as well as control esiRNA (targeted against the GFP), were purchased from Sigma–Aldrich (Saint Louis, MO, U.S.A.). Transfection was performed using the JetPrime (Polyplus, Illkirch-Graffenstaden, France) transfection kit. HEK293T cells were seeded at a density of 2 × 104 cells per 450 µl in each well of 24-well plates. EsiRNA was diluted in 48 µl Jetprime buffer; then, 2 µl of JetPrime reagent was added to the mixture, which was incubated at 20°C for 15 min. The final mixture was added to culture plate wells, resulting in a final esiRNA concentration of 40 and 60 nM. Cells were collected after 72 h incubation, and protein and total RNA samples were prepared.

Generation of HEK293 ACACA MTS knockout-derived cell lines

CRISPR/Cas9-mediated genome editing was used to generate a HEK293-derived cell line with an inactivated ACC1 MTS (Supplementary Figure S4A). We used a commercial, custom-manufactured CRISPR/Cas9 genome-editing construct based on the pSpCas9(BB)-2 A-GFP vector containing an sgRNA expression construct (Sigma–Aldrich, Saint Louis, MO, U.S.A.) designed to target the exon encoding the ACC1 MTS. The supplier used their own proprietary algorithms to choose a construct with maximum specificity. One single construct was proposed by Sigma–Aldrich due to the short target area (ACACA exon 2/ENSE00003512420, 47 nucleotides: GTCTTTCTGGAAGTGGATATCTACTCAGACAGTAAGAATTATAAGAG).

HEK293T cells were transfected with this CRISPR/Cas9 genome-editing construct. Cells expressing the CRISPR/Cas9 constructs were identified by fluorescence-activated cell sorting, and candidate clones for successful genome editing were identified by a T7 endonuclease assay (see Supplementary Experimental). PCR products of the edited region were cloned in a vector, and several clones were subsequently sequenced to identify the mutations in the knockout cell line (Supplementary Figure S4). Sequence alignments were generated using Clustal Omega [25].

Real-time quantitative PCR

Acaca shRNA KD

Total RNA isolation (RNeasy Kit, Qiagen, Hilden, Germany) and first-strand cDNA synthesis (RevertAid First Strand cDNA Synthesis Kit, Fermentas/ThermoFisher Waltham, MA, U.S.A.) were done according to the manufacturers’ protocols. Total RNA (1 µg) treated with DNase I was used for the reverse transcription reaction; 1/25 of the synthesized DNA was used for the 25 µl PCR. The real-time PCR was carried out using a TaqMan Gene Expression Assay (Invitrogen, Carlsbad, CA, U.S.A.) specific for the mouse Acaca (Mm01304277_m1) according to the manufacturer's protocol. As the internal control, the TaqMan Gene Expression Assay (Invitrogen, Carlsbad, CA, U.S.A.) specific for Eukaryotic 18s rRNA (4333960F) was used. To obtain a relative quantification, the ΔΔCt method was used in the analysis. Student's t-test was employed as a parametric statistical tool for the analysis. The standard deviation was calculated to measure the amount of dispersion of the dataset.

ACSF3 esiRNA KD

Total RNA was isolated using the RNeasy Mini kit (Qiagen, Hilden, Germany) and treated with DNase (Qiagen, Hilden, Germany) to avoid genomic DNA contamination. Total RNA yield was determined using a Nano-drop® ND-1000 Spectrophotometer. Total RNA (150 ng) was reverse-transcribed using the RevertAid First Strand cDNA Synthesis Kit according to the manufacturer's instructions. The resulting cDNA was stored in aliquots at −70°C. Real-time PCR was performed with Taqman (Thermo Fisher Scientific, Waltham, MA, U.S.A.) probes for human ACSF3 and as endogenous control Actin B was used. A 7500 real-time PCR system (Applied Biosystems, Foster City, CA, U.S.A.) was employed for the run and data analysis. Relative quantification and statistical analysis were carried out as described above.

Western blotting

Acaca shRNA KD

Protein extracts from the cells were prepared using the Proteojet Mammalian Cell Lysis Reagent (Thermo Fisher Scientific, Waltham, MA, U.S.A.), according to the manufacturer's protocol. Mitochondria were extracted from the cell culture using the Mitochondria Isolation Kit for Cultured Cells (Abcam, MA, U.S.A.) according to the manufacturer's protocol. The proteins were separated by SDS–polyacrylamide gel electrophoresis. The proteins were transferred to the membrane nitrocellulose membrane (Bio-Rad TURBO 0.2 µm, Bio-Rad, Hercules, CA, U.S.A.) using the Trans-Blot Turbo (Bio-Rad, Hercules, CA, U.S.A.) and Sigma–Aldrich (Saint Louis, MO, U.S.A.) Blocking buffer was used for blocking of the membrane. Rabbit monoclonal ACC1-specific antibody (Millipore, Billerica, MA, U.S.A.) was used at a 1 : 1000 dilution. For rabbit polyclonal anti-lipoic acid antibody (Calbiochem, La Jolla, CA, U.S.A.), a 1 : 3000 dilution was used. The anti-β-actin antibody (Novus Biologicals, Littleton, CO, U.S.A.) was used at a 1 : 5000 dilution. The secondary antibodies peroxidase-conjugated goat anti-rabbit IgG (Sigma–Aldrich, Saint Louis, MO, U.S.A.) and goat anti-mouse IgG (Bio-Rad, Hercules, CA, U.S.A.) were used at 1 : 10 000 dilutions. Chemiluminescence detection was performed using the Immun-Star WesternC Kit (Bio-Rad, Hercules, CA, U.S.A.). The PageRuler™ Prestained protein Ladder (Fermentas/Thermo Fisher Scientific, Waltham, MA, U.S.A.) was used for the estimation of the relative molecular mass of proteins. The signal from the western blot was detected using the ChemiDoc XRS molecular imager (Bio-Rad, Hercules, CA, U.S.A.) and analyzed with the Image Lab 3.0.1 software (Bio-Rad, Hercules, CA, U.S.A.) where applicable. The volume of the LA (lipoic acid) and actin signals was normalized using the global background subtraction method.

Mecr KD

SDS–PAGE for the western blot using the Mecr antibody was done using 12% Tris–glycine gel. SDS–PAGE for the western blot using the lipoic acid antibody was done using 10% Tris–glycine gel. Proteins on the gels were transferred on a nitrocellulose membrane using the Trans-Blot Turbo system (Bio-Rad, Hercules, CA, U.S.A.). Membranes were blocked with Casein blocking buffer solution (Sigma–Aldrich, Saint Louis, MI, U.S.A.). Membranes were probed using the following antibodies: for probing Mecr, as a primary antibody rabbit polyclonal anti-Mecr-IgG (Proteintech, Chicago, U.S.A.) and as a secondary antibody goat anti-rabbit-IgG-HRP conjugate (Hercules, CA, U.S.A.) were used. For probing lipoic acid, as a primary antibody rabbit anti-LA-IgG (Calbiochem, La Jolla, CA, U.S.A.) and as a secondary goat anti-rabbit-IgG-HRP conjugate (Hercules, CA, U.S.A.) were used. For probing Dlat, as a primary antibody mouse polyclonal anti-Dlat-IgG (Novus Biologicals, Littleton, CO, U.S.A.) and as a secondary antibody goat anti-mouse-IgG-HRP conjugate (Madison, WI, U.S.A.) were used. In all cases, Porin (VDAC) was probed as a loading control, using as a primary antibody mouse anti-Porin-IgG (Abcam, MA, U.S.A.) and as a secondary antibody goat anti-mouse-IgG-HRP conjugate. Signal was detected using the Clarity ECL Western Blotting Substrate (Bio-Rad, Hercules, CA, U.S.A.), and the blots were imaged with the Chemidoc XRS camera system (Bio-Rad, Hercules, CA, U.S.A.).

ACACA MTS KO

Mitochondrial extracts (20 µg) were loaded in a 10% SDS–PAGE gel (Bio-Rad, Hercules, CA, U.S.A.), transferred to a nitrocellulose membrane (Bio-Rad, Hercules, CA, U.S.A.) and blocked with 5% fat-free dried milk powder dissolved in Tris-buffered saline containing 0.1% Tween 20. After blocking, the membrane was incubated with the primary antibody, then washed in Tris-buffered saline containing 0.1% Tween 20 and incubated with the secondary antibody. A first screen to select potential knockout cell lines was done by using an anti-lipoic acid primary antibody (same Calbiochem monoclonal antibody as above, now sold by Merk Millipore, Billerica, MA, U.S.A.). Cell lines that lacked or displayed a reduced detectable signal for lipoylated proteins were subjected to a secondary immunoblot screen using antisera that recognize the DLST E2 subunit of α-ketoglutarate dehydrogenase (anti-dihydrolipoamide S-succinyltransferase E2, mouse polyclonal; Novus Biologicals, Littleton, CO, U.S.A.), or the DLAT protein (anti-pyruvate dehydrogenase E2, mouse polyclonal, Novus Biologicals, Littleton, CA, U.S.A.). Goat anti-rabbit-HRP (1 : 10000, Abcam, MA, U.S.A.) and anti-mouse-HRP (1 : 3333, Madison, WI, U.S.A.) were used as secondary antibodies. Clarity™ Western ECL Blotting Substrates (Bio-Rad, Hercules, CA, U.S.A.) were used according to manufacturer's recommendations. Quantification of band intensities was performed using the Image Lab software. To calculate the percent change of lipoylation in the ΔMTS-ACC1 mutant, three biological repeats (mitochondrial extract from separate cell cultivations) were analyzed. To obtain comparable results, ratios of the lipoylation signal over the corresponding β-actin loading control were determined, and thus, normalized level of wild-type (WT) lipoic acid over WT β-actin was defined as 100% for each individual experiment. Percent lipoic acid in the sample was determined as the fraction of [LA/β-actin] from the ΔMTS-ACC1 mutant compared with [LA/β-actin] of wild-type cells. A two-tailed Student's t-test was used to determine the P-value.

Fluorescence microscopy

Co-localization study

HEK293 cells were seeded on a coverslip in a 12-well plate containing DMEM, high glucose complemented with 1 mM uridine, 5 mM sodium pyruvate, 10% (v/v) fetal calf serum, 1× essential amino acids and 1× penicillin–streptomycin at 37°C under an atmosphere of 5% CO2 and 95% air. After the cells had attached, transfection was performed using FuGENE HD Transfection (Promega, Madison, WI, U.S.A.) according to the manufacturer's instructions. To detect the mitochondrial network, cells were also transfected with the MtDsRED plasmid (pDsRED-Mito, Clontech/Takara Bio, Mountain View, CA, U.S.A.). After 24 h incubation, cells were fixed with 4% PFA, fluorescence microscopic pictures were taken with a Zeiss LSM700 confocal microscope and 3D images were created from stacks processing analysis. The overlap coefficient (Manders) and correlation coefficient (Pearson) were collected from the 3D representation.

Acaca shRNA KD

For the study of the mitochondrial morphology of mouse Acaca KD cell lines, a ZEISS (ZEISS, Oberkochen, Germany) Observer-Z1confocal microscope with a ZEISS LSM700 unit was used. The images were obtained with a plan APOCHROMAT 60× objective oil lens using the ZEN 2009 software. The MitoTracker Red CMXRos (Invitrogen, Carlsbad, CA, U.S.A.) was used to stain the mitochondria following the manufacturer's protocol.

Mecr KD

Subconfluent cell lines were seeded on glass coverslips on cell culture plates and grown overnight. MitoTracker Red CMXRos (Invitrogen, Carlsbad, CA, U.S.A.) diluted in DMEM at a concentration of 100 nM was used to stain the mitochondria of the cells. Cells were fixed using 4% PFA solution and the glass coverslip was mounted on the microscopy slide using Vectashield Mounting Medium with DAPI (Vector Laboratories, Burlingame, CA, U.S.A.). Mitochondrial morphology was imaged using a Zeiss LSM 700 confocal microscope.

ACACA MTS KO

HEK293 cells (WT and ACACA MTS KO) were seeded on a coverslip in a 6-well plate containing DMEM, high glucose complemented with 1 mM uridine, 5 mM sodium pyruvate, 10% (v/v) fetal calf serum, 1× essential amino acids and 1× penicillin–streptomycin at 37°C under an atmosphere of 5% CO2 and 95% air. After the cells attached, transfection was performed using FuGENE HD Transfection (Promega, Madison, WI, U.S.A.) according to the manufacturer's instructions. After 48 h incubation, cells were fixed with 4% PFA and fluorescence microscopic pictures were taken with a Zeiss LSM700 confocal microscope.

Results

The human ACACA-001 encoded ACC1 protein isoform harbors a functional N-terminal MTS

An intra-mitochondrial mammalian ACC has not been described to date. Inspired by our previous finding that cytosolic and mitochondrial ACCs in yeast were originally dually localized but expressed from one gene [13], we investigated transcript variants of ACCs in the human genome listed in ENSEMBL [26]. The database lists four transcript splicing variants encoding full-size ACCs that vary at the predicted N-termini (Figure 1A). The probabilities of these N-terminal sequences serving as mitochondrial import signals were assessed using MitoProt II [18], TargetP [19] and iPSORT [20]. Our results indicate that the ACC1 protein isoform 1, produced from the human transcript ACACA-001, is predicted to be mitochondrially localized with high probability by all three algorithms (Figure 1B).

We generated a construct that expresses a fusion of the first 37 amino acid residues of ACC1 isoform 1 to GFP in HEK293 cells in order to investigate if this N-terminal sequence constitutes a functional MTS. Figure 1A depicts the ACC1 amino acid sequence fused to GFP in gray highlights. Fluorescence microscopy of cells expressing a control GFP construct without the predicted N-terminal MTS exhibited diffuse green fluorescence and negative correlation with the MtDsrRED mitochondrial marker signal (Figure 1C,E). In contrast, when GFP appended with the predicted MTS of ACC1 isoform 1 was expressed in these cells, the green fluorescent signal showed a significantly higher overlap and a positive correlation with the MtDsRED mitochondrial marker (Figure 1D,E).

Mitochondrial targeting sequences of mammalian ACACAs are highly conserved

To investigate how well the N-terminal MTS is conserved among species, we conducted an N-BLAST search against vertebrate protein sequences using the predicted MTS of the human ACC1 isoform 1 protein. Our search returned ACC1 proteins from a wide range of mammalian species. Investigation of the N-terminal sequences of these ACC1 orthologs using the MitoProtII, Target P and iPSORT program indicated that many of these candidate proteins were predicted to harbor sequences with high probability of constituting N-terminal MTSs. A list with selected homologs can be found in Table 1, among them also a putative alligator sequence as an example of a non-mammalian vertebrate.

Table 1
Conservation of putative mitochondrial ACC isoforms in vertebrates

The predicted MTS of the human ACC1 isoform 1 protein was used in an N-BLAST search against vertebrate protein sequences. The table lists of 10 selected homologs predicted to localize to mitochondria with high probability. Mitochondrial localization prediction was carried out with MitoprotII, Target P and iPSORT mitochondrial import prediction software. Where the programs predicted a targeting sequence cleavage site, we have included this information (number of amino acids from initiation methionine) in brackets.

Species Definition GenBank accession no. Mitoprot (CS) TargetP (CS) iPSORT mitochondrial prediction 
Homo sapiens Human Acetyl-CoA carboxylase 1 isoform 1 NP_942131.1 0.9930 (13) 0.868 (22) Yes 
Mus musculus Mouse Acetyl-coenzyme A carboxylase α, partial CAF02251.1 0.1062 (NP) 0.075 (NP) No 
Mus musculus Mouse Predicted: acetyl-CoA carboxylase 1 isoform X2 XP_006532016.1 0.7977 (28) 0.873 (19) Yes 
Macaca mulatta Rhesus monkey Acetyl-CoA carboxylase 1 precursor NP_001253707.1 0.9855 (27) 0.815 (12) Yes 
Bos taurus Cattle Acetyl-CoA-carboxylase α, partial CAX51837.1 0.9828 (13) 0.715 (12) Yes 
Ovis aries Sheep Acetyl-CoA carboxylase-α CAD92090.1 0.9849 (35) 0.783 (34) Yes 
Pan troglodytes Chimpanzee Predicted: acetyl-coenzyme A carboxylase α XP_511428.4 0.9922 (13) 0.884 (22) Yes 
Oryctolagus cuniculus Rabbit Predicted: acetyl-CoA carboxylase 1 isoform X3 XP_008269381.1 0.9817 (27) 0.820 (12) Yes 
Ailuropoda melanoleuca Giant panda Predicted: acetyl-CoA carboxylase 1 isoform X1 XP_019661138.1 0.9846 (27) 0.831 (12) Yes 
Alligator mississippiensis Alligator Predicted: acetyl-CoA carboxylase 1 isoform X5 XP_014463555.2 0.9275 (43) 0.930 (16) Yes 
Species Definition GenBank accession no. Mitoprot (CS) TargetP (CS) iPSORT mitochondrial prediction 
Homo sapiens Human Acetyl-CoA carboxylase 1 isoform 1 NP_942131.1 0.9930 (13) 0.868 (22) Yes 
Mus musculus Mouse Acetyl-coenzyme A carboxylase α, partial CAF02251.1 0.1062 (NP) 0.075 (NP) No 
Mus musculus Mouse Predicted: acetyl-CoA carboxylase 1 isoform X2 XP_006532016.1 0.7977 (28) 0.873 (19) Yes 
Macaca mulatta Rhesus monkey Acetyl-CoA carboxylase 1 precursor NP_001253707.1 0.9855 (27) 0.815 (12) Yes 
Bos taurus Cattle Acetyl-CoA-carboxylase α, partial CAX51837.1 0.9828 (13) 0.715 (12) Yes 
Ovis aries Sheep Acetyl-CoA carboxylase-α CAD92090.1 0.9849 (35) 0.783 (34) Yes 
Pan troglodytes Chimpanzee Predicted: acetyl-coenzyme A carboxylase α XP_511428.4 0.9922 (13) 0.884 (22) Yes 
Oryctolagus cuniculus Rabbit Predicted: acetyl-CoA carboxylase 1 isoform X3 XP_008269381.1 0.9817 (27) 0.820 (12) Yes 
Ailuropoda melanoleuca Giant panda Predicted: acetyl-CoA carboxylase 1 isoform X1 XP_019661138.1 0.9846 (27) 0.831 (12) Yes 
Alligator mississippiensis Alligator Predicted: acetyl-CoA carboxylase 1 isoform X5 XP_014463555.2 0.9275 (43) 0.930 (16) Yes 

Several mRNA variants of Acaca are documented for mouse in the ENSEMBL database. Acaca-005, an incomplete transcript identified by 5′-RACE, is predicted to encode a variant of Acc1 with an extended N-terminus compared with the Acaca-001- or Acaca-201 encoded full-size Acc1 (Supplementary Figure S1A). Mitochondrial localization prediction analysis of the N-terminal extension with the three independent algorithms indicated a high probability of constituting an MTS with an MMP cleavage site at amino acid 28 (Supplementary Figure S1A, cleaved sequence underlined), leaving the Acc1 enzyme intact after removal of the signal peptide. A predicted protein isoform containing this N-terminus and with high probability of mitochondrial localization is listed in Genbank as XP_006532016.1 (Table 1).

A specific KD of the putative mitochondrial isoforms of mouse Acaca results in mitochondrial defects

A KD targeting total Acaca transcripts can be predicted to exert pleiotropic negative effects on the cellular metabolism, because Acc1 plays essential roles in many important processes. Hence, we designed constructs for retrovirus-mediated permanent KDs [22] expressing shRNAs specifically targeted for RNAi toward the section of these transcripts predicted to encode the putative MTS of Acc1. In addition, we generated constructs targeting all Acaca transcripts in mouse NIH 3T3 cells for comparison. Two KD constructs specific for the transcript encoding the putative mitochondrial isoform and three KD constructs targeting all full-size mRNA species were tested. Only one of the former (referred to M1) and one of the latter (A1) resulted in the desired KD effect. Cells expressing either one of the effective shRNA constructs grew slower than those infected with a virus harboring the empty control construct, not expressing shRNA.

The transcript levels of Acaca in the total Acc1 KD cell lines (A1) were reduced to less than half of the level in controls, while cell lines knocked down specifically for the putative mitochondrial isoform of Acc1(M1) showed reduction in expression levels by ∼15% compared with the control (Supplementary Figure S2). Congruently, western blotting confirmed that there was a clear reduction in Acc1 protein levels in the A1 KD cell line, while no clear decrease in Acc1 could be detected in the M1 KD cells (Figure 2A, upper panel).

The Acaca KD in mouse cells affects protein lipoylation and mitochondrial morphology.

Figure 2.
The Acaca KD in mouse cells affects protein lipoylation and mitochondrial morphology.

(A) Western blotting analysis of total cell extract from mouse NIH3T3 KD cell lines. The shRNAi targeting of total Acc1 transcript (shRNA A1) but not the specific targeting of the transcript variant encoding the putative mitochondrial Acc1 isoform resulted in reduction of Acc1 protein levels in whole cell extracts (upper panel: anti-Acc1; loading control: β-actin). Protein lipoylation of the Dlat E2 subunit of pyruvate dehydrogenase as well as Dlst of α-ketoglutarate dehydrogenase was reduced in the Acaca KD, targeting either the transcript variant encoding cytosolic Acc1 or exclusively the variant encoding the mitochondrial isoform (lower panel, anti-lipoic acid; loading control: β-actin). The M2 lane between the M1 and A1 samples corresponds to the unsuccessful M2 KD cell line (see Results section and Supplementary Figure S1) for which the analysis of the cell extract displayed a lipoylation pattern identical with that of the WT sample. Dlat: dihydrolipoamide S-acetyltransferase, E2 component of the pyruvate dehydrogenase complex; Dlst: dihydrolipoamide S-succinyltransferase, E2 component of two oxoglutarate complex. Control: cells infected with virus carrying the empty vector. (B) Mitochondrial fragmentation upon KD of Acaca in mouse NIH3T3 KD cells. The control cells exhibit mostly reticular mitochondria, while the mitochondrial isoform-specific knockdown cells (shRNA M1) contain mostly fragmented mitochondria. Cells expressing the total Acaca KD shRNA (shRNA A1) have an intermediate mitochondrial morphology phenotype. Mitochondria are visualized with MitoTracker Red CMX Ros. Left panels are enlarged sections of the panels on the right. KD constructs used for the study are indicated on top of the left-side panels. Cells were imaged using confocal microscopy. Scale bars are 10 µm.

Figure 2.
The Acaca KD in mouse cells affects protein lipoylation and mitochondrial morphology.

(A) Western blotting analysis of total cell extract from mouse NIH3T3 KD cell lines. The shRNAi targeting of total Acc1 transcript (shRNA A1) but not the specific targeting of the transcript variant encoding the putative mitochondrial Acc1 isoform resulted in reduction of Acc1 protein levels in whole cell extracts (upper panel: anti-Acc1; loading control: β-actin). Protein lipoylation of the Dlat E2 subunit of pyruvate dehydrogenase as well as Dlst of α-ketoglutarate dehydrogenase was reduced in the Acaca KD, targeting either the transcript variant encoding cytosolic Acc1 or exclusively the variant encoding the mitochondrial isoform (lower panel, anti-lipoic acid; loading control: β-actin). The M2 lane between the M1 and A1 samples corresponds to the unsuccessful M2 KD cell line (see Results section and Supplementary Figure S1) for which the analysis of the cell extract displayed a lipoylation pattern identical with that of the WT sample. Dlat: dihydrolipoamide S-acetyltransferase, E2 component of the pyruvate dehydrogenase complex; Dlst: dihydrolipoamide S-succinyltransferase, E2 component of two oxoglutarate complex. Control: cells infected with virus carrying the empty vector. (B) Mitochondrial fragmentation upon KD of Acaca in mouse NIH3T3 KD cells. The control cells exhibit mostly reticular mitochondria, while the mitochondrial isoform-specific knockdown cells (shRNA M1) contain mostly fragmented mitochondria. Cells expressing the total Acaca KD shRNA (shRNA A1) have an intermediate mitochondrial morphology phenotype. Mitochondria are visualized with MitoTracker Red CMX Ros. Left panels are enlarged sections of the panels on the right. KD constructs used for the study are indicated on top of the left-side panels. Cells were imaged using confocal microscopy. Scale bars are 10 µm.

The yeast mitochondrial Hfa1 ACC provides the malonyl-CoA extender unit to mtFAS, which produces the octanoic acid precursor of lipoic acid. To investigate if mtFAS is affected by the Acaca KDs in mouse cells, we analyzed lipoic acid contents of the Acaca KD cell lines. Western blots of protein extracts from KD and control cell lines were probed with anti-lipoic acid antibody. Initially, extracts analyzed from the slow-growing M1 and A1 cell lines displayed a reduction or disappearance, respectively, of the lipoylation signal of a specific subpopulation of the lipoylated Dlat subunit of the PDH complex (Supplementary Figure S3A). This phenotype was reminiscent of a Dlat lipoylation defect in mtFAS that we observed upon KD of expression of mouse Mecr, encoding mitochondrial enoyl-CoA/ACP reductase of mtFAS (Supplementary Figure S3B, left lower panel). Reduction of the lipoylated signal was accompanied by a concomitant reduction of total Dlat (Supplementary Figure S3C). Later analysis, using a different lot of the lipoic acid antibody, did not reveal such a clear separation into two different subspecies, but confirmed an overall reduction of lipoylation in both A1 and M1 cell lines compared with the control (Figure 2A, lower panel).

We also investigated if inactivation of the mitochondrial ACC isoform would have consequences for the mitochondrial morphology in the KD cell lines. Compared with the control cell line, which displayed mostly tubular mitochondrial structure, KD of the mitochondrial isoform of Acc1 resulted in mitochondrial fragmentation (Figure 2B, central panels). Like our lipoylation analysis results, these phenotypes correlate well with the data we obtained in Mecr KD studies (Supplementary Figure S3).

CRISPR/Cas9-mediated inactivation of the predicted MTS of human ACC1 leads to lipoylation defects and altered mitochondrial morphology

Our mouse bioinformatics data and the results of the Acaca KD experiments supported our hypothesis of the presence of mitochondrial isoforms of Acc1 in mammals. Owing to the rather short target area of the extra exon of the mouse RNA encoding the putative mitochondrial Acc1 isoform, we were unable to generate an shRNA construct that would have allowed us to independently verify an effect of a KD of this splicing variant on lipoylation in mouse. To strengthen our data, we set out to investigate if results similar to the data as obtained by experiments with mouse cultured cells could be independently obtained using a heterologous mammalian cell model system. As our studies on the functionality of the putative MTS harbored by isoform 1 of human ACC1 indicated that this protein variant is likely to localize to mitochondria, we decided to again turn to human cells to address the question on a mitochondrial function of ACC1, employing CRISPR/Cas9 editing [27]. We purchased a CRISPR/Cas9 editing construct designed by a commercial supplier (Sigma–Aldrich) based on the bioinformatics and experimental evidence, we had previously generated. Using this construct, we edited the genome of the HEK293 cell line to harbor mutations that were predicted to inactivate the putative MTS encoded by the ACACA transcript isoform 1. The clones were initially screened with the T7 endonuclease I assay for successful editing, and the presence of inactivating mutations was confirmed by sequencing (Supplementary Figure S4). As shown in the alignment of the alleles found in the ΔMTS-ACC1 clone to the WT sequence, we were able to obtain ACACA alleles edited to harbor indel mutations in the sequences encoding the predicted MTS of ACACA. We found a 1 bp insertion in one of the alleles and a 2 bp deletion in the other ACACA allele, in both cases leading to reading frameshifts and introduction of early stop codons (Supplementary Figure S4B,C). The sequence encoding the part of ACC1 shared by all isoforms, representing the active enzyme, is predicted to remain intact in these alleles (Supplementary Figure S4C).

We first examined the level of ACC1 from total cell extracts by using anti-ACC1 serum. Similar to our observations in the mouse M1 KD cell line described above, we were unable to detect any obvious change of ACC1 levels in the ΔMTS-ACC1 compared with WT. In contrast, we were able to observe a robust decrease in the level of lipoylated DLAT, the pyruvate dehydrogenase E2 subunit, to 35.9 (±28)% (n = 3, P = 0.019) compared to WT in crude mitochondrial extracts. Concomitantly, overall DLAT protein levels appeared reduced when we probed with an antibody specific for this E2 subunit of PDH (Figure 3A), similar to the result we obtained using the Mecr KD mouse cell line (Supplementary Figure S3A). As in our analysis of the mouse Acc1 isoform 1 KD cells, we also investigated the effect of inactivation of the predicted MTS of human ACC1 on mitochondrial morphology. To visualize the mitochondria, WT and ΔMTS-ACC1 HEK293 cells were transfected with a plasmid that target the Green Fluorescence Protein to the outer membrane of the mitochondrial compartment. Our analysis of confocal microscopic image of the fluorescent mitochondrial network in the ΔMTS-ACC1 cells indicated a severe fragmentation phenotype (Figure 3B).

Effect of CRISPR/Cas9-mediated inactivation of the ACC1 MTS in human cells on protein lipoylation and mitochondrial fragmentation and lipoylation defect by ACSF3 KD.

Figure 3.
Effect of CRISPR/Cas9-mediated inactivation of the ACC1 MTS in human cells on protein lipoylation and mitochondrial fragmentation and lipoylation defect by ACSF3 KD.

(A) Protein lipoylation of the DLAT subunit was reduced in the ΔMTS-ACC1 cell line. Protein analysis by western blotting of HEK293 total cell extracts for the detection with anti-ACC1 (upper panel), with a corresponding loading control probing for β-actin. Analysis of mitochondrial extracts with anti-LA (lipoic acid, middle panel) or anti-DLAT (dihydrolipoamide S-acetyltransferase, component E2 of the multi-enzyme PDH complex, lower panel) antiserum revealed a reduction of both lipoylation and DLAT protein. Anti-VDAC1 (voltage-dependent anion-selective channel 1, antibodies were used as a loading control. (B) Fragmentation of mitochondria in the ΔMTS-ACC1 human cell line. Cells were transfected with a plasmid expressing and targeting the GFP (green fluorescent protein) to the mitochondrial outer membrane. On the top panel, fluorescence microscopic picture of the HEK293 WT control cells exhibiting reticular mitochondrial structure, while ΔMTS-ACC1 cells depicted on the lower panels show mitochondrial fragmentation. Zoomed pictures are shown on the left panels. Cells were imaged using confocal microscopy. Scale bars are 10 µm. (C) The DLAT subunit protein lipoylation defect was exacerbated by the ACSF3 KD in the ΔMTS-ACC1 cell line. Protein analysis by western blotting of the HEK293 cell extracts was carried out with anti-LA (lipoic acid, upper panel) antibody. Anti-ACTIN (lower panel) antibody was used as a loading control.

Figure 3.
Effect of CRISPR/Cas9-mediated inactivation of the ACC1 MTS in human cells on protein lipoylation and mitochondrial fragmentation and lipoylation defect by ACSF3 KD.

(A) Protein lipoylation of the DLAT subunit was reduced in the ΔMTS-ACC1 cell line. Protein analysis by western blotting of HEK293 total cell extracts for the detection with anti-ACC1 (upper panel), with a corresponding loading control probing for β-actin. Analysis of mitochondrial extracts with anti-LA (lipoic acid, middle panel) or anti-DLAT (dihydrolipoamide S-acetyltransferase, component E2 of the multi-enzyme PDH complex, lower panel) antiserum revealed a reduction of both lipoylation and DLAT protein. Anti-VDAC1 (voltage-dependent anion-selective channel 1, antibodies were used as a loading control. (B) Fragmentation of mitochondria in the ΔMTS-ACC1 human cell line. Cells were transfected with a plasmid expressing and targeting the GFP (green fluorescent protein) to the mitochondrial outer membrane. On the top panel, fluorescence microscopic picture of the HEK293 WT control cells exhibiting reticular mitochondrial structure, while ΔMTS-ACC1 cells depicted on the lower panels show mitochondrial fragmentation. Zoomed pictures are shown on the left panels. Cells were imaged using confocal microscopy. Scale bars are 10 µm. (C) The DLAT subunit protein lipoylation defect was exacerbated by the ACSF3 KD in the ΔMTS-ACC1 cell line. Protein analysis by western blotting of the HEK293 cell extracts was carried out with anti-LA (lipoic acid, upper panel) antibody. Anti-ACTIN (lower panel) antibody was used as a loading control.

ACC1 and ACSF3 operate in concert in protein lipoylation

Complete inactivation of the ACC1 MTS in HEK293 cells only leads to partial loss of lipoylation in the ΔMTS-ACC1 cells. It has been suggested previously that ACSF3, a mitochondrial malonyl-CoA synthetase, is acting as the malonyl-CoA provider for mtFAS [9]. Despite the appropriate location and function of this enzyme, however, an RNAi-mediated KD of ACSF3 in HEK293 cells line did not result in a protein lipoylation defect [9]. To test the hypothesis that ACSF3 acts as a malonyl-CoA producer in a parallel fashion to ACC1, we investigated the effect of ACSF3 depletion on protein lipoylation in the ΔMTS-ACC1 cell line. Using 40 nM of ACSF3 siRNA, expression of ACSF3 was reduced to 60% compared with the control after 72 h (Supplementary Figure S5). Similar to the results obtained by Witkowski et al. [28], this reduction of ACSF3 expression did not result in any detectable loss of lipoylation. However, when we treated the cells with 60 nM ACSF3 siRNA, reducing ACSF3 transcript levels to 40% of WT expression, we were able to observe a mild hypolipoylation phenotype (Figure 3C). Moreover, when ΔMTS-ACC1 cells were exposed to ACSF3 siRNA at 40 nM concentration, we were not able to detect further reduction of DLAT lipoylation, while lipoyl-DLAT was reduced to barely detectable levels in ΔMTS-ACC1 cells treated with 60 nM siRNA specific for ACSF3. These results are consistent with the model that ACSF3 and the mitochondrial isoform of ACC1 operate in tandem to supply malonyl-CoA to human mtFAS.

Discussion

The role of malonyl-CoA as a regulator and the co-ordinator of fatty acid metabolism have been long acknowledged. Apart from acting as the basic building blocks for FAS, the cytosolic concentration of malonyl-CoA is the cellular indicator of availability of acetyl-CoA in this compartment, which controls the activity of mitochondrial carnitine palmitoyl transferase and therefore, by extension, the β-oxidation activity in mitochondria. We have previously suggested that mtFAS may act as an intramitochondrial acetyl-CoA sensor [10]. Our work presented here provides evidence that mammalian ACC1 and ACSF3 collaborate in the production of malonyl-CoA required for mtFAS, indicating an important role of ACC1 in acetyl-CoA sensing also in mitochondria, with malonyl-CoA acting as a master signal of energy metabolism.

Our data provide clear evidence that the conserved N-terminal extension of ACC1 isoform 1 in humans acts as a functional mitochondrial targeting signal. Our further analysis demonstrates that a specific KD of the corresponding isoform in mouse results in lipoylation and mitochondrial morphology defects congruent with the phenotypes of a KD of the transcript of Mecr of mtFAS.

Furthermore, we show that inactivation of this sequence by genomic editing of the targeting signal-encoding sequence in human cells produces congruent results. All this evidence points to a role of ACC1 isoform 1 in mtFAS and adds the last missing piece to the puzzle of mtFAS components in mammals. The reason why ACC1 had not been previously reported to be localized in mitochondria is probably due to the phenomenon called ‘eclipsed distribution’ [29], where the localization of a small subpopulation of a protein is not readily detected because the overwhelming majority is found elsewhere in the cell. Such a case has been well documented for aconitase in yeast [29], although with reversed roles compared with mammalian ACC1. The bulk of aconitase, which has long been regarded as a mitochondrial marker protein, is localized to mitochondria where it participates in the TCA cycle. However, a minute amount of this enzyme is present in the cytosol, where it serves in the glyoxylate cycle. There are a few examples of mitochondrial isoforms of proteins in mammals by alternative splicing-generated by the addition of a sequence encoding an N-terminal MTS [30], but no systematic investigation has been made to address this question. While proteomics data do not currently support the notion that protein isoform generation by alternative splicing is a major contributor to proteome complexity [31], it cannot be ruled out that low-abundance alternative protein variants may escape grand-scale analysis approaches.

In contrast with a previous report on the effects of a KD of expression of ACP in a human cell culture model, which presented data implying that the DLAT protein turns over independently from the lipoic acid [16], we observed a decrease in Dlat protein concomitantly with the decrease in lipoylation in both of our model systems. Feng et al., however, used a transient KD approach for their study. In contrast, our approaches produced a permanent KD mouse cell line and a human cell line that carries persisting genomic mutations in activating the MTS of ACC1. We therefore surmise that the disappearance of the Dlat/DLAT proteins in our studies is due to prolonged delipoylation of the protein, eventually leading to its disappearance.

It is worth noting that, like mammals, apparently also many fungi also produce both the cytosolic and the mitochondrial matrix versions of ACC from one gene [13]. As an interesting twist in the fungal case, however, these lower eukaryotes seem to prefer generating the mitochondrial isoform by translation from a non-AUG initiation codon upstream of the canonical AUG that initiates production of the cytosolic form [13], rather than using a mechanism involving an alternative exon. This may be due to the fact that the extremely streamlined fungal genomes generally contain very few introns, and using alternative translation initiation mechanism is the most frugal approach to making alternate versions of proteins that are required in multiple locations.

As an additional intriguing deviation in mammals from the yeast model, animal ACC1 apparently works in concert with a malonyl-CoA synthetase — ACSF3 — in the generation of malonyl-CoA. Why this apparently redundant function exists may remain a subject of speculation. A fundamental difference between yeast and mammalian mitochondria is the lack of β-oxidation in mitochondria of the former. Owing to the alternative source of acetyl-CoA directly produced in the mammalian organelle, an additional layer of control of mtFAS may be required.

In spite of this alternative source of malonyl-CoA produced by the ACSF3 enzyme, our data indicate that the general mechanism of malonyl-CoA generation for mtFAS is conserved from yeast to man. Poignantly, while monocotyledon plants have been shown to use ACCs in malonyl-CoA synthesis for mtFAS [32], dicots have recently been reported to employ a mitochondrial isoform of a malonyl-CoA synthetase named AEE13 for mtFAS/protein lipoylation [33].

ACC1 (ACACA) is listed in Mitocarta as a potential mitochondrial disease gene [34]. As we provide evidence for a role of a mitochondrial isoform of ACC1 in mtFAS, mutant variants of ACC1 with compromised enzymatic function or dysfunction of the mitochondrial targeting signal, for example due to a splicing defect, may indeed lead to a mitochondrial disease phenotype. It should therefore be a matter of investigation if such deficiencies could be contributing to neurodegenerative disease phenotypes similar to the recently described MECR dysfunction in MEPAN [11]. Several cases of ACSF3 disorder have been reported, and patients display symptoms that partially overlap with the symptoms of MEPAN patients [35].

The presence of ACC1 in mitochondria also adds a new aspect to the proposed role of this enzyme in cancer. The presence of ACC1 is absolutely required for breast cancer cell survival [36], and translation of ACACA has been reported to be up-regulated in breast cancer cells [37]. In line with these observations, the inactive, phosphorylated form of ACC1 (P-ACC1) [38] has been shown to be associated with longer survival in lung cancer patients [39]. A link of the proposed role of ACC1 in breast cancer exists via the BRCA1 oncogene, which has been reported to interact with P-ACC via its BRCT1 domains [40,41]. The active phosphorylated form of BRCA1 is mainly localized to the nucleus and mitochondria [42], where ACC1 had not been found previously. As our data clearly suggest that also the ACC1 protein is present in this compartment, an interaction between BRCA1 and ACC1 in mitochondria may be a key component to the observed regulatory function of BRCA1 in energy metabolism [43]. It has been shown that a low energy status can suppress the malignant phenotype of certain cancers [44]. Such a down-regulation of energy metabolism may be achieved by attenuating mtFAS via maintaining an inactive form of mitochondrial ACC1 by BRCA1 binding, a regulatory mechanism that may be disturbed in cancer cells harboring BRCA1 mutations. Our work may therefore add a new facet to the proposed ‘Metabolic Syndrome of (Breast) cancer’ [45].

Abbreviations

     
  • 5′-RACE

    5′ rapid amplification of cDNA ends

  •  
  • ACC

    acetyl-coA carboxylase

  •  
  • ACP

    acyl carrier protein

  •  
  • DMEM

    Dulbecco's Modified Eagle Medium

  •  
  • eGFP

    enhanced green fluorescent protein

  •  
  • FAS

    fatty acid synthesis

  •  
  • FBS

    fetal bovine serum

  •  
  • HEK

    human embryonic kidney

  •  
  • KD

    knockdown

  •  
  • LA

    lipoic acid

  •  
  • MECR

    mitochondrial enoyl-CoA/ACP reductase

  •  
  • MEPAN

    mitochondrial enoyl-CoA reductase protein-associated neurodegeneration

  •  
  • mtFAS

    mitochondrial fatty acid synthesis

  •  
  • MTS

    mitochondrial targeting signal

  •  
  • ORF

    open reading frame

  •  
  • PFA

    paraformaldehyde

  •  
  • WT

    wild type

Author Contribution

A.J.K. conceived the study. G.M., F.S., J.M.K. and A.J.M. performed in vitro and in vivo experiments. G.M., F.S., J.M.K. and A.J.K. did data analysis and wrote the manuscript.

Funding

This work was supported by the Academy of Finland, the Sigrid Juselius Foundation, the Biocenter Oulu and AFM-Téléthon.

Acknowledgments

We thank Prof. Kalervo Hiltunen for support, advice and critical reading of the manuscript, Dr Aki Manninen for the help with the retrovirus-mediated knockdown procedure, Hamayun Khan for his contribution in the set-up of the initial mouse cell knockdown experiments and Leila Polus for technical support.

Competing Interests

The Authors declare that there are no competing interests associated with the manuscript.

References

References
1
Tong
,
L.
and
Harwood
, Jr,
H.J.
(
2006
)
Acetyl-coenzyme A carboxylases: versatile targets for drug discovery
.
J. Cell. Biochem.
99
,
1476
1488
2
Barber
,
M.C.
,
Price
,
N.T.
and
Travers
,
M.T.
(
2005
)
Structure and regulation of acetyl-CoA carboxylase genes of metazoa
.
Biochim. Biophys. Acta Mol. Cell Biol. Lipids
1733
,
1
28
3
Abu-Elheiga
,
L.
,
Brinkley
,
W.R.
,
Zhong
,
L.
,
Chirala
,
S.S.
,
Woldegiorgis
,
G.
and
Wakil
,
S.J.
(
2000
)
The subcellular localization of acetyl-CoA carboxylase 2
.
Proc. Natl Acad. Sci. U.S.A.
97
,
1444
1449
4
Saggerson
,
D.
(
2008
)
Malonyl-CoA, a key signaling molecule in mammalian cells
.
Annu. Rev. Nutr.
28
,
253
272
5
White
,
S.W.
,
Zheng
,
J.
,
Zhang
,
Y.-M.
and
Rock
,
C.O.
(
2005
)
The structural biology of type II fatty acid biosynthesis
.
Annu. Rev. Biochem.
74
,
791
831
6
Kastaniotis
,
A.J.
,
Autio
,
K.J.
,
Kerätär
,
J.M.
,
Monteuuis
,
G.
,
Mäkelä
,
A.M.
,
Nair
,
R.R.
et al. 
(
2017
)
Mitochondrial fatty acid synthesis, fatty acids and mitochondrial physiology
.
Biochim. Biophys. Acta Mol. Cell Biol. Lipids
1862
,
39
48
7
Mayr
,
J.A.
,
Feichtinger
,
R.G.
,
Tort
,
F.
,
Ribes
,
A.
and
Sperl
,
W.
(
2014
)
Lipoic acid biosynthesis defects
.
J. Inherit. Metab. Dis.
37
,
553
563
8
Wada
,
H.
,
Shintani
,
D.
and
Ohlrogge
,
J.
(
1997
)
Why do mitochondria synthesize fatty acids? Evidence for involvement in lipoic acid production
.
Proc. Natl Acad. Sci. U.S.A.
94
,
1591
1596
9
Witkowski
,
A.
,
Joshi
,
A.K.
and
Smith
,
S.
(
2007
)
Coupling of the de novo fatty acid biosynthesis and lipoylation pathways in mammalian mitochondria
.
J. Biol. Chem.
282
,
14178
14185
10
Kursu
,
V.A.S.
,
Pietikäinen
,
L.P.
,
Fontanesi
,
F.
,
Aaltonen
,
M.J.
,
Suomi
,
F.
,
Raghavan Nair
,
R.
et al. 
(
2013
)
Defects in mitochondrial fatty acid synthesis result in failure of multiple aspects of mitochondrial biogenesis in Saccharomyces cerevisiae
.
Mol. Microbiol.
90
,
824
840
11
Heimer
,
G.
,
Kerätär
,
J.M.
,
Riley
,
L.G.
,
Balasubramaniam
,
S.
,
Eyal
,
E.
,
Pietikäinen
,
L.P.
et al. 
(
2016
)
MECR mutations cause childhood-onset dystonia and optic atrophy, a mitochondrial fatty acid synthesis disorder
.
Am. J. Hum. Genet.
99
,
1229
1244
12
Hoja
,
U.
,
Marthol
,
S.
,
Hofmann
,
J.
,
Stegner
,
S.
,
Schulz
,
R.
,
Meier
,
S.
et al. 
(
2004
)
HFA1 encoding an organelle-specific acetyl-CoA carboxylase controls mitochondrial fatty acid synthesis in Saccharomyces cerevisiae
.
J. Biol. Chem.
279
,
21779
21786
13
Suomi
,
F.
,
Menger
,
K.E.
,
Monteuuis
,
G.
,
Naumann
,
U.
,
Kursu
,
V.A.S.
,
Shvetsova
,
A.
et al. 
(
2014
)
Expression and evolution of the non-canonically translated yeast mitochondrial acetyl-CoA carboxylase Hfa1p
.
PLoS ONE
9
,
e114738
14
Halenz
,
D.R.
,
Feng
,
J.Y.
,
Hegre
,
C.S.
and
Lane
,
M.D.
(
1962
)
Some enzymic properties of mitochondrial propionyl carboxylase
.
J. Biol. Chem.
237
,
2140
2147
PMID:
[PubMed]
15
Chen
,
C.
,
Han
,
X.
,
Zou
,
X.
,
Li
,
Y.
,
Yang
,
L.
,
Cao
,
K.
et al. 
(
2014
)
4-methylene-2-octyl-5-oxotetrahydrofuran-3-carboxylic acid (C75), an inhibitor of fatty-acid synthase, suppresses the mitochondrial fatty acid synthesis pathway and impairs mitochondrial function
.
J. Biol. Chem.
289
,
17184
17194
16
Feng
,
D.
,
Witkowski
,
A.
and
Smith
,
S.
(
2009
)
Down-regulation of mitochondrial acyl carrier protein in mammalian cells compromises protein lipoylation and respiratory complex I and results in cell death
.
J. Biol. Chem.
284
,
11436
11445
17
Li
,
W.
,
Cowley
,
A.
,
Uludag
,
M.
,
Gur
,
T.
,
McWilliam
,
H.
,
Squizzato
,
S.
et al. 
(
2015
)
The EMBL-EBI bioinformatics web and programmatic tools framework
.
Nucleic Acids Res.
43
,
W580
W584
18
Claros
,
M.G.
and
Vincens
,
P.
(
1996
)
Computational method to predict mitochondrially imported proteins and their targeting sequences
.
Eur. J. Biochem.
241
,
779
786
19
Emanuelsson
,
O.
,
Nielsen
,
H.
,
Brunak
,
S.
and
von Heijne
,
G.
(
2000
)
Predicting subcellular localization of proteins based on their N-terminal amino acid sequence
.
J. Mol. Biol.
300
,
1005
1016
20
Bannai
,
H.
,
Tamada
,
Y.
,
Maruyama
,
O.
,
Nakai
,
K.
and
Miyano
,
S.
(
2002
)
Extensive feature detection of N-terminal protein sorting signals
.
Bioinformatics
18
,
298
305
21
Nair
,
R.R.
,
Kerätär
,
J.M.
,
Autio
,
K.J.
,
Masud
,
A.J.
,
Finnilä
,
M.A.J.
,
Autio-Harmainen
,
H.I.
et al. 
(
2017
)
Genetic modifications of Mecr reveal a role for mitochondrial 2-enoyl-CoA/ACP reductase in placental development in mice
.
Hum. Mol. Genet.
26
,
2104
2117
22
Schuck
,
S.
,
Manninen
,
A.
,
Honsho
,
M.
,
Fullekrug
,
J.
and
Simons
,
K.
(
2004
)
Generation of single and double knockdowns in polarized epithelial cells by retrovirus-mediated RNA interference
.
Proc. Natl Acad. Sci. U.S.A.
101
,
4912
4917
23
Brummelkamp
,
T.R.
,
Bernards
,
R.
and
Agami
,
R.
(
2002
)
A system for stable expression of short interfering RNAs in mammalian cells
.
Science
296
,
550
553
24
Travers
,
M.T.
,
Cambot
,
M.
,
Kennedy
,
H.T.
,
Lenoir
,
G.M.
,
Barber
,
M.C.
and
Joulin
,
V.
(
2005
)
Asymmetric expression of transcripts derived from the shared promoter between the divergently oriented ACACA and TADA2L genes
.
Genomics
85
,
71
84
25
Goujon
,
M.
,
McWilliam
,
H.
,
Li
,
W.
,
Valentin
,
F.
,
Squizzato
,
S.
,
Paern
,
J.
et al. 
(
2010
)
A new bioinformatics analysis tools framework at EMBL-EBI
.
Nucleic Acids Res.
38
(
Web Server issue
),
W695
W699
26
Aken
,
B.L.
,
Ayling
,
S.
,
Barrell
,
D.
,
Clarke
,
L.
,
Curwen
,
V.
,
Fairley
,
S.
et al. 
(
2016
)
The Ensembl gene annotation system
.
Database
2016
,
baw093
27
Ran
,
F.A.
,
Hsu
,
P.D.
,
Wright
,
J.
,
Agarwala
,
V.
,
Scott
,
D.A.
and
Zhang
,
F.
(
2013
)
Genome engineering using the CRISPR-Cas9 system
.
Nat. Protoc.
8
,
2281
2308
28
Witkowski
,
A.
,
Thweatt
,
J.
and
Smith
,
S.
(
2011
)
Mammalian ACSF3 protein is a malonyl-CoA synthetase that supplies the chain extender units for mitochondrial fatty acid synthesis
.
J. Biol. Chem.
286
,
33729
33736
29
Regev-Rudzki
,
N.
,
Karniely
,
S.
,
Ben-Haim
,
N.N.
and
Pines
,
O.
(
2005
)
Yeast aconitase in two locations and two metabolic pathways: seeing small amounts is believing
.
Mol. Biol. Cell
16
,
4163
4171
30
Pereira
,
K.D.
,
Tamborlin
,
L.
,
Meneguello
,
L.
,
de Proença
,
A.R.G.
,
Almeida
,
I.C.d.P.A.
,
Lourenco
,
R.F.
et al. 
(
2016
)
Alternative start codon connects eIF5A to mitochondria
.
J. Cell. Physiol.
231
,
2682
2689
31
Tress
,
M.L.
,
Abascal
,
F.
and
Valencia
,
A.
(
2017
)
Alternative splicing may not be the key to proteome complexity
.
Trends Biochem. Sci.
42
,
98
110
32
Focke
,
M.
,
Gieringer
,
E.
,
Schwan
,
S.
,
Jänsch
,
L.
,
Binder
,
S.
and
Braun
,
H.-P.
(
2003
)
Fatty acid biosynthesis in mitochondria of grasses: malonyl-coenzyme A is generated by a mitochondrial-localized acetyl-coenzyme A carboxylase
.
Plant Physiol.
133
,
875
884
33
Guan
,
X.
and
Nikolau
,
B.J.
(
2016
)
AAE13 encodes a dual-localized malonyl-CoA synthetase that is crucial for mitochondrial fatty acid biosynthesis
.
Plant J.
85
,
581
593
34
Calvo
,
S.E.
,
Clauser
,
K.R.
and
Mootha
,
V.K.
(
2016
)
Mitocarta2.0: an updated inventory of mammalian mitochondrial proteins
.
Nucleic Acids Res.
44
,
D1251
D1257
35
Sloan
,
J.L.
,
Johnston
,
J.J.
,
Manoli
,
I.
,
Chandler
,
R.J.
,
Krause
,
C.
,
Carrillo-Carrasco
,
N.
et al. 
(
2011
)
Exome sequencing identifies ACSF3 as a cause of combined malonic and methylmalonic aciduria
.
Nat. Genet.
43
,
883
886
36
Chajès
,
V.
,
Cambot
,
M.
,
Moreau
,
K.
,
Lenoir
,
G.M.
and
Joulin
,
V.
(
2006
)
Acetyl-CoA Carboxylase α Is Essential to Breast Cancer Cell Survival
.
Cancer Res.
66
,
5287
5294
37
Yoon
,
S.
,
Lee
,
M.-Y.
,
Park
,
S.W.
,
Moon
,
J.-S.
,
Koh
,
Y.-K.
,
Ahn
,
Y.-H.
et al. 
(
2007
)
Up-regulation of acetyl-CoA carboxylase α and fatty acid synthase by human epidermal growth factor receptor 2 at the translational level in breast cancer cells
.
J. Biol. Chem.
282
,
26122
26131
38
Kudo
,
N.
,
Barr
,
A.J.
,
Barr
,
R.L.
,
Desai
,
S.
and
Lopaschuk
,
G.D.
(
1995
)
High rates of fatty acid oxidation during reperfusion of ischemic hearts are associated with a decrease in malonyl-CoA levels due to an increase in 5′-AMP-activated protein kinase inhibition of acetyl-CoA carboxylase
.
J. Biol. Chem.
270
,
17513
17520
39
Conde
,
E.
,
Suarez-Gauthier
,
A.
,
García-García
,
E.
,
Lopez-Rios
,
F.
,
Lopez-Encuentra
,
A.
,
García-Lujan
,
R.
et al. 
(
2007
)
Specific pattern of LKB1 and phospho-acetyl-CoA carboxylase protein immunostaining in human normal tissues and lung carcinomas
.
Hum. Pathol.
38
,
1351
1360
40
Magnard
,
C.
,
Bachelier
,
R.
,
Vincent
,
A.
,
Jaquinod
,
M.
,
Kieffer
,
S.
,
Lenoir
,
G.M.
et al. 
(
2002
)
BRCA1 interacts with acetyl-CoA carboxylase through its tandem of BRCT domains
.
Oncogene
21
,
6729
6739
41
Ray
,
H.
,
Moreau
,
K.
,
Dizin
,
E.
,
Callebaut
,
I.
and
Venezia
,
N.D.
(
2006
)
ACCA phosphopeptide recognition by the BRCT repeats of BRCA1
.
J. Mol. Biol.
359
,
973
982
42
Coene
,
E.D.
,
Hollinshead
,
M.S.
,
Waeytens
,
A.A.T.
,
Schelfhout
,
V.R.J.
,
Eechaute
,
W.P.
,
Shaw
,
M.K.
et al. 
(
2005
)
Phosphorylated BRCA1 is predominantly located in the nucleus and mitochondria
.
Mol. Biol. Cell
16
,
997
1010
43
Jackson
,
K.C.
,
Gidlund
,
E.-K.
,
Norrbom
,
J.
,
Valencia
,
A.P.
,
Thomson
,
D.M.
,
Schuh
,
R.A.
et al. 
(
2014
)
BRCA1 is a novel regulator of metabolic function in skeletal muscle
.
J. Lipid Res.
55
,
668
680
44
Swinnen
,
J.V.
,
Beckers
,
A.
,
Brusselmans
,
K.
,
Organe
,
S.
,
Segers
,
J.
,
Timmermans
,
L.
et al. 
(
2005
)
Mimicry of a cellular low energy status blocks tumor cell anabolism and suppresses the malignant phenotype
.
Cancer Res.
65
,
2441
2448
45
Brunet
,
J.
,
Vazquez-Martin
,
A.
,
Colomer
,
R.
,
Graña-Suarez
,
B.
,
Martin-Castillo
,
B.
and
Menendez
,
J.A.
(
2008
)
BRCA1 and acetyl-CoA carboxylase: the metabolic syndrome of breast cancer
.
Mol. Carcinog.
47
,
157
163

Author notes

*

These authors contributed equally to this work.

Present address: Institute of Biotechnology, University of Helsinki, 00290 Helsinki, Finland.

Supplementary data